Abstract
Dear Editor,
We read with interest the randomized trial by He et al. 1 evaluating continuous versus intermittent anticoagulation during endoscopic variceal treatment in cirrhotic patients with acute portal vein thrombosis (PVT). The authors demonstrated the safety and efficacy of continuous anticoagulation and found this regimen to be associated with lower rebleeding rates and further improved survival at 12 months. We would wish to highlight several concerns with the study.
The most important limitation of this trial, in our opinion, is its study design, which weakens its conclusions, particularly with protocol transparency, outcome switching, and the distinction between exploratory analyses and prospectively designed randomized comparisons. The study is presented as a randomized controlled trial (RCT) but actually represents a subgroup analysis of another RCT. We appreciate that the authors acknowledge this in the methodology and limitations sections, but it is not reflected elsewhere (title, abstract, and introduction). These analyses are consequently merely exploratory and cannot substitute an adequately powered RCT. Furthermore, there is no mention of a pre-specified subgroup analysis in the originally published trial protocol. The primary endpoint of the trial changed from comparing two injection techniques of sclerotherapy over a duration of 3 months to estimating the rebleeding risk with anticoagulation over 12 months without any published substantial protocol amendment. More than 60% of patients with a history of variceal bleed included in the original trial had acute PVT, which seems implausible. Moreover, authors should provide details about the timeline of protocol amendment and the beginning of the new RCT, as we believe the protocol amendment must have taken place when more than 50% of patients in the original RCT were identified as having PVT. The question for the authors is how these patients were managed and what the lead time bias was with regard to the anticoagulation strategy (continuous vs intermittent) before the trial underwent protocol amendment. A review of the original and amended protocols would be useful, and we hope the authors can provide this.
Several other points do not align with the current management principles in practice and hence limit the generalizability of results. First, the strategy of interrupting anticoagulation for 7 days following every banding. Second, the initial dose of 4000 units twice daily of low-molecular-weight heparin, irrespective of body weight, cannot be regarded as therapeutic anticoagulation. Third, because of the points mentioned above, the data detailing time to PVT recanalization in the two groups becomes important; however, it is unfortunately not presented in this study. The apparent benefit in late rebleeding and mortality has no mechanistic explanation in the absence of PVT outcomes, especially when the early rebleeding rate (the effect of anticoagulation) remains similar. The effect sizes for rebleeding and mortality, while significant, are imprecise with wide confidence intervals and the upper limit nearly reaching unity.
The authors should be commended for addressing this research question, which is undoubtedly important for the entire Hepatology community, but unfortunately, the above concerns limit the interpretation of their results and conclusions.
